<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title><![CDATA[Kimi 长文阅读强，但知识库问答仍要权限过滤]]></title><description><![CDATA[<p dir="auto">我们想用 Kimi 做公司知识库，直接把部门文档塞给模型。权限怎么处理？模型会自己区分吗？</p>
]]></description><link>https://localaihub.com/topic/88/kimi-长文阅读强-但知识库问答仍要权限过滤</link><generator>RSS for Node</generator><lastBuildDate>Wed, 03 Jun 2026 19:44:24 GMT</lastBuildDate><atom:link href="https://localaihub.com/topic/88.rss" rel="self" type="application/rss+xml"/><pubDate>Tue, 05 May 2026 08:46:00 GMT</pubDate><ttl>60</ttl><item><title><![CDATA[Reply to Kimi 长文阅读强，但知识库问答仍要权限过滤 on Wed, 06 May 2026 09:58:00 GMT]]></title><description><![CDATA[<p dir="auto">收到。先改成按用户权限检索，再把可见文档送模型。</p>
]]></description><link>https://localaihub.com/post/603</link><guid isPermaLink="true">https://localaihub.com/post/603</guid><dc:creator><![CDATA[北岛不是岛]]></dc:creator><pubDate>Wed, 06 May 2026 09:58:00 GMT</pubDate></item><item><title><![CDATA[Reply to Kimi 长文阅读强，但知识库问答仍要权限过滤 on Wed, 06 May 2026 07:19:00 GMT]]></title><description><![CDATA[<p dir="auto">前端权限是体验，不是安全边界。知识库问答的安全边界在后端检索、上下文构造、日志和审计。</p>
]]></description><link>https://localaihub.com/post/602</link><guid isPermaLink="true">https://localaihub.com/post/602</guid><dc:creator><![CDATA[林小北]]></dc:creator><pubDate>Wed, 06 May 2026 07:19:00 GMT</pubDate></item><item><title><![CDATA[Reply to Kimi 长文阅读强，但知识库问答仍要权限过滤 on Wed, 06 May 2026 05:14:00 GMT]]></title><description><![CDATA[<p dir="auto">我们原来只做前端权限，看来不够。</p>
]]></description><link>https://localaihub.com/post/601</link><guid isPermaLink="true">https://localaihub.com/post/601</guid><dc:creator><![CDATA[北岛不是岛]]></dc:creator><pubDate>Wed, 06 May 2026 05:14:00 GMT</pubDate></item><item><title><![CDATA[Reply to Kimi 长文阅读强，但知识库问答仍要权限过滤 on Wed, 06 May 2026 03:59:00 GMT]]></title><description><![CDATA[<p dir="auto">还有日志。prompt 里有敏感资料，日志系统别明文长期存。</p>
]]></description><link>https://localaihub.com/post/600</link><guid isPermaLink="true">https://localaihub.com/post/600</guid><dc:creator><![CDATA[今天不想改配置]]></dc:creator><pubDate>Wed, 06 May 2026 03:59:00 GMT</pubDate></item><item><title><![CDATA[Reply to Kimi 长文阅读强，但知识库问答仍要权限过滤 on Wed, 06 May 2026 02:09:00 GMT]]></title><description><![CDATA[<p dir="auto">面向用户可以简洁显示“依据：售后政策 2026-04”。内部审计记录完整 chunk id。</p>
]]></description><link>https://localaihub.com/post/599</link><guid isPermaLink="true">https://localaihub.com/post/599</guid><dc:creator><![CDATA[leaf_1997]]></dc:creator><pubDate>Wed, 06 May 2026 02:09:00 GMT</pubDate></item><item><title><![CDATA[Reply to Kimi 长文阅读强，但知识库问答仍要权限过滤 on Tue, 05 May 2026 23:18:00 GMT]]></title><description><![CDATA[<p dir="auto">但引用太多会不会用户烦？</p>
]]></description><link>https://localaihub.com/post/598</link><guid isPermaLink="true">https://localaihub.com/post/598</guid><dc:creator><![CDATA[葡萄冰]]></dc:creator><pubDate>Tue, 05 May 2026 23:18:00 GMT</pubDate></item><item><title><![CDATA[Reply to Kimi 长文阅读强，但知识库问答仍要权限过滤 on Tue, 05 May 2026 21:21:00 GMT]]></title><description><![CDATA[<p dir="auto">回复里也要引用来源。用户看到答案来自哪份文档，能帮发现权限和过期问题。</p>
]]></description><link>https://localaihub.com/post/597</link><guid isPermaLink="true">https://localaihub.com/post/597</guid><dc:creator><![CDATA[阿宁]]></dc:creator><pubDate>Tue, 05 May 2026 21:21:00 GMT</pubDate></item><item><title><![CDATA[Reply to Kimi 长文阅读强，但知识库问答仍要权限过滤 on Tue, 05 May 2026 20:05:00 GMT]]></title><description><![CDATA[<p dir="auto">检索前过滤一遍，检索后再过滤一遍。权限字段要进索引元数据，不要只在前端菜单控制。</p>
]]></description><link>https://localaihub.com/post/596</link><guid isPermaLink="true">https://localaihub.com/post/596</guid><dc:creator><![CDATA[zeroOne]]></dc:creator><pubDate>Tue, 05 May 2026 20:05:00 GMT</pubDate></item><item><title><![CDATA[Reply to Kimi 长文阅读强，但知识库问答仍要权限过滤 on Tue, 05 May 2026 18:48:00 GMT]]></title><description><![CDATA[<p dir="auto">如果检索系统误召回了别人部门文档呢？</p>
]]></description><link>https://localaihub.com/post/595</link><guid isPermaLink="true">https://localaihub.com/post/595</guid><dc:creator><![CDATA[小周]]></dc:creator><pubDate>Tue, 05 May 2026 18:48:00 GMT</pubDate></item><item><title><![CDATA[Reply to Kimi 长文阅读强，但知识库问答仍要权限过滤 on Tue, 05 May 2026 17:19:00 GMT]]></title><description><![CDATA[<p dir="auto">Kimi 的长文能力适合用户有权限的一组长材料，比如一个项目空间内的会议记录和需求文档。</p>
]]></description><link>https://localaihub.com/post/594</link><guid isPermaLink="true">https://localaihub.com/post/594</guid><dc:creator><![CDATA[陈一]]></dc:creator><pubDate>Tue, 05 May 2026 17:19:00 GMT</pubDate></item><item><title><![CDATA[Reply to Kimi 长文阅读强，但知识库问答仍要权限过滤 on Tue, 05 May 2026 14:57:00 GMT]]></title><description><![CDATA[<p dir="auto">长上下文容易让人偷懒，把“候选资料”全塞进去。企业知识库最危险就是这个偷懒。</p>
]]></description><link>https://localaihub.com/post/593</link><guid isPermaLink="true">https://localaihub.com/post/593</guid><dc:creator><![CDATA[nora]]></dc:creator><pubDate>Tue, 05 May 2026 14:57:00 GMT</pubDate></item><item><title><![CDATA[Reply to Kimi 长文阅读强，但知识库问答仍要权限过滤 on Tue, 05 May 2026 13:52:00 GMT]]></title><description><![CDATA[<p dir="auto">对。进了上下文就等于暴露给模型推理。即使模型不直接引用，也可能间接泄露。</p>
]]></description><link>https://localaihub.com/post/592</link><guid isPermaLink="true">https://localaihub.com/post/592</guid><dc:creator><![CDATA[rootless]]></dc:creator><pubDate>Tue, 05 May 2026 13:52:00 GMT</pubDate></item><item><title><![CDATA[Reply to Kimi 长文阅读强，但知识库问答仍要权限过滤 on Tue, 05 May 2026 12:12:00 GMT]]></title><description><![CDATA[<p dir="auto">也就是说用户没权限看的文档，连 prompt 都不能进？</p>
]]></description><link>https://localaihub.com/post/591</link><guid isPermaLink="true">https://localaihub.com/post/591</guid><dc:creator><![CDATA[小傅]]></dc:creator><pubDate>Tue, 05 May 2026 12:12:00 GMT</pubDate></item><item><title><![CDATA[Reply to Kimi 长文阅读强，但知识库问答仍要权限过滤 on Tue, 05 May 2026 11:49:00 GMT]]></title><description><![CDATA[<p dir="auto">权限必须在检索和上下文组装前处理，不能指望模型替你保密。</p>
]]></description><link>https://localaihub.com/post/590</link><guid isPermaLink="true">https://localaihub.com/post/590</guid><dc:creator><![CDATA[林小北]]></dc:creator><pubDate>Tue, 05 May 2026 11:49:00 GMT</pubDate></item></channel></rss>